ANALYTICAL CHEMISTRY
1578 perature coefficients appear to be in qualitative agreement with the relative chemical reactivity of toluene, the xylenes, and benzene. coYcLusloYs
The temperature coefficient of absorbance in the ultraviolet spectral region is a significant factor for certain aromatic hydrocarbons and some chelated aliphatic oxygenated compounds. Precise quantitative analyses of mixtures t h a t contain a n y of these substances require control of the temperature of the sample, or empirical corrections for ohserved deviations in temperature. For the determination of other compounds noted herein, the variation of absorbance with temperature is usually negligible compared with the normal manipulatory errors.
LITERATURE CITED
(1) .iriiold, L. B., Jr., and Kistiakowsky. G . B.. J . Am. Chem. Soc., 54, 1713 (1932). (2) Ayres, G. H., and Tuffly, B. L., A B . ~ LCHEY., , 23, 788 (1951). (3) Bastian, R., Ibid., 25,259 (1953). (4) Beale, R. N., and Roe, E. 11. F., J . Sci. Instr., 28, 109 (1951). (5) Bowden, F. P., and Snow, C. P., Proc. Rog. Soc. ( L o n d o n ) , 115B,261 (1934). (6) Haupt, G. W., J . Rescarch Natl. Bur. Sfandards, 48, 414 (1952) (RP 2331). (7) Kortum, G., and Halban, -4.v., 2. physik. Chena., 170A, 212 f1934).
(8) JIukerj;, B. K., Bhattacharji, A. K., and Dhar, W. R., J . PhUs. Chem., 32, 1834 (1938). (9) Sheppard, S.E., and Brigham, H. R., J . Am. Chem. Soc., 66. 380 (1944). (10) Sklar, A. I,., J . Chem. Phys., 5 , 669 (1937). R L G L IE\ D for review J l a r c h 15, 1934.
Accepted July 1, 1954
Determination of Sodium and Potassium Oxides by Flame Photometry In Portland Cement Raw Materials and Mixtures and Similar Silicates C. L. FORD Analytical Laboratories, Research and Development Division, Portland Cement Association,
A method was needed whereby acid-insoluble silicates, such as are used in the manufacture of portland cement, could be brought rapidly into solution for determination of sodium and potassium oxides by flame photometry techniques in common use for cement. The desired results were attained by sintering and extracting the material by the classical J. Lawrence Smith method, then removing the unknown amount of calcium in the extract and substituting a known amount (plus hydrochloric acid) approximating that found in the standard solutions used in the method for cement. The alkali contents obtained by flame photometry were in good agreement with those obtained by comparable gravimetric analyses of the same silicates. I n using the new method it is not necessary to prepare special solutions or otherwise alter the flame photometer technique used for portland cement. The time requirements are much less than for gravimetric determination of alkalies in silicates.
T
HE increasing interest in the alltali content of portland cement has led to an increasing demand for determination of the sodium and potassium oxide content of cement, and the raw materials from which it is made, and an increasing need for 1111proved analytical methods. This need has been met, in the case of cement, bv an .4STM tentative method ( 2 ) using flame photometry. Similar short methods (not ASTlI) for determining the alkali content of cement raw mateiials and tnixtures, or similar silicates, were incomplete or not designed for the cement testing laboratory. The study reported herein was undertahen to develop for the cement chemist a n accurate method of analysis of raw materials by flanir photometry which would require the minimum of special preparation and manipulation. G E b E R 4 L COY SIDER4TION S
Before any niaterial can be analj zed by the flame photometer, the desired constituents must be in solution. T h e nearly coniplete solubility of cement in dilute hydrochloric acid mahes possible the relativelv simple procedure of the tentative ASTLI flame
33 W e s t Grand Ave., Chicago 10,111.
photometer method ( 2 ) . The partial inPo1ubility of most cement raw materials (except high-grade limestones) requires some other method of dissolving the alkalies. .Z majority of investigators ( 3 , 4 , 6 - 8 , 1 0 ) of photometric procedures have used some modification of the well-known J. Lawrence Smith sintering procedure to decompose the sample, followed by est,racting the sinter cake n-ith Imter. [This procedure is used in the BSTM referee method ( I ) , in which the alkalies are finally determined gravimetrically.] The water extract contains the alkalies and varying amourit,s of calcium derived largely from the calcium carbonate used in the sintering procedure. The aniount varies with the way the sinter cake is extracted. The author’s principal efforts have been directed toward developing a method which would be accurate and would permit the cement chemist to use the same standard solutions that are used for cement testing bj- flame phot,ometry. Cement laboratories using the flame phot,omet,erare likely to itock such standard solutions; hence if solutions of the ran- materials could be prepared to contain similar amounts of calcium. same solutions could be used for standardization of the instrument in testing both cements and raw materials. This mag be done by removing the unknown amount of calcium present in the sinter cake extract and replacing it with a known amount approximately the same as that present in the cement solutions used for test according t,o the hSTN photometric method. PROCEDURE
Preparation and Analysis of Sample. Weigh 0.5 to 5.0 grams of the finely pulverized raw material and mix it with 0.5 to 1.0 gram of ammonium chloride by grinding the two together in a n agate or mullite mortar for 3 minutes (for choice of sample size see the discussion of procedure). Add 1 grams of calcium carbonate to the mixture and continue the grinding until t,he ingredients are thoroughly mixed (7 to 10 minutes). Place about 0.5 gram of calcium carbonate on the bottom of a 40-ml. platinum crucible. add the mixture t o the crucible, and compact the mixture wit’h a “muehroom” glass rod. Place about 0.5 gram of calcium carbonate in the mortar and grind for about 2 minutes to “rinse” the mortar and pestle. Distribute t,his calcium carbonate on the mixture in the crucible. Mount the crucible through a hole in a n asbestos board about 0.3 cm. thick, so bhat the rrucihle projects about 1 cm. above the hoard. Cover the crucible with a well-fit,ting cover and set on i t another platinum cruciiile or a 50-ml. beaker that is kept. filled
V O L U M E 2 6 , N O , 10, O C T O B E R 1 9 5 4 with water through the entire period of ignition. Heat the crucible a t a very low temperature a t first and gradually raise t,he temperature within 1 hour to a bright red heat. At no time should the temperature be raised so fast that white ammonium chloride fumes can be seen issuing from the crucible. Finally, ignite the crucible so that the temperature of the contents will be maintained at. 900" t o 1000" C. for 1 hour. Cool the crurible, and by inverting into a 250-ml. beaker and tapping, remove as much of the sinter as possible. Moisten the sinter in the beaker with 2 to 3 ml. of water, and add hot water until the crucible is nearly filled. Cover the beaker with a watch glass and place under an infrared lamp. As the sinter in the beaker absorbs the water, add more, so that there is always a 2or 3-ml. excess on the bottom. Continue until disintegration of the sinter is complete. Remove the beaker from the heat sourw. transfer the solution in the crucible to the beaker, wruh the crucible with a policeman, and rinse into the beaker. If an appreciable amount of the sinter cannot be removed from the crucible, place the crucihle and contents in the beaker. By addition of xYcer bring the contents of the beaker to about 100 ml., or if the beaker contains the crucible, add enough water to cover the crucible. Digest the mixture below the boiling point for 4 hours or overnight. Break any lumps with a pestle or mushroom glass rod. Boil the mixtui,e for 1 minute, allow the residue and decant the clear solution through an 11-em. S o . 31 Whatman ( o r equivalent,) filter paper into a 600-ml. beaker. -4dd 30 ml. of water to the 250-nil. beaker, bring the mixture to a boil, allow the residue to settle. and decant t,he filt'rate through the Same filter paper. Perform this addition of m t e r and decantation three more times. Finally, \I-ash the residue into the filter paper, rinse the beaker, and wash the paper and residue 8 to 10 times with hot water. Cover t h r 600-nil. beaker containing the filtrate and washings, and boil the solution rapidly to a volume of about 50 ml. Cool the solution to near room temperature. Add 3 to 4 grams of ammonium carbonate (powdered), bring to a boil, and continue boiling for a few seconds only. Filter the solution through a 9-em. No. 30 Whatnian (or equivalent) filter paper into a 250-ml. beaker. Transfer the residue to the paper with the aid of cold ammonium carbonate solution (10 grams per liter), and wash the paper and contents eight times with the ammonium carbonate solution. Evaaorate the filtrate and washings to a volume of about 50 nil. without loss of material through effervescence of escaping gases or bumping. This may be done rapidly by placing the covered beaker on a low-temperature hot plate under an infrared heat lamp. Add 1 drop of methyl red indicator (2 grams per liter of ethyl alcohol) to the solution and a few drops of concentrated hydrochloric acid to neutralize. Cool to room temperature and transfer t h e solution into a volumetric flask of suitable size containing 10 ml. of calcium chloride solution (112.5 grams of calcium carbonate and 500 ml. of concentrated hydrochloric acid per liter of a ueous solutions) for each 100 ml. of volume of the flask. (Therkask size should he such t h a t the concentration of sodium oxide or potassium oxide does not exceed 100 p.p.m,) Dilute to the mark with water and shake well. Determine the sodium and potassium oxide content ok the solution ( 8 ) . Calculation of Results. Using the recorded average of meter readings for sodium oxide and potassium oxide in the unknown sumple, read the concentration in part's per million of each oxide from its calibration curve. If the solution of the sample contained 1.0 gram of sample per 100 ml. of solution, 1 p.p.m. represents 0.01 TCof alkali oxide.
Table I .
1579 Otherwise calculate as follows:
cx
1-
% S a 2 0 or IizO = If- x 10,000 where C = p.p.m. of S a 2 0 or KIO (from curve) 1- = volume of flask containing the sample, nil. TI' = weight of sample, grams 10,000 = p.p.m. of standard solution equivalent to 1.0% of Na20 or K20 DISCUSSION OF PROCEDURE
The size of sample depends on bot,h t'he :alcium and the alkali content of the mat'erial. I n the analysis of rocks, according to Hillebrand and Lundell (8),the rat,io of sample to calcium carbonate should he about 1 to 8, or 0.5 to 4.0 grams. Therefore, in the analyses of silicates low in calcium, 0.5-gram samples should be t.al;eri. Hon-ever. because both cements and cement raw niixes contain large aniountr of calcium with relative11 less insoluble silicates t o be decomposed, samples up to 5.0 grams may he used. The alkali content must be considered also in deciding on both the sample size and amount of final dilution, since the maximum concentration of either sodium or potassium oxide must not exceed 100 p.p,m.. the ma.,imum concentration of the standard solutions. I n general, the largest possible sample should be used for greatest accuracy. The filtrate obtained after treatment x i t h ammonium carbonate is very low in calcium. Hence the addition of calcium chloride stock solution as described above gives solutions of the raw mat,erial comparable in calcium and acid content to the standard alkali solutions. The flame photometer technique then becomes similar to that for acid solutions of cement. The final filtrate is reduced to about 50 ml., not only to permit transfer to flasks as small as 100 ml., but, t o drive off most of the ammonium carbonate. If the ammonium carbonat,e is not removed. on neutralization an excessive amount of ammonium chloride will be formed and has been found to interfere with the sodium determination. The final volume of the solution depends on the alkali content. The concentration must be kept below 100 p.p.m. For example, if a 1.0-gram sample has been used and the sodium oxide content is expect,ed to be between 1.0 and 2.0%) the solution should be made up to 200 ml. I n case the approximate alkali content is not known, it is better to prepare a small volume, then, if necessary, further dilute an aliquot by the addition of an appropriate amount, of calcium chloride stock solution. EXPERIMEBTAL WORK
Five samples each of raw mixtures and kiln dusts were analyzed for alkalies by both the .4STL\I gravimetric and proposed flame photoniet.er methods. A sample of cement, (KO. 11668, also identified as PCA check sample 1) \vas analyzed by the 1 S T M gravimetric method, the -4STM photomet,er method, and the proposed flame photometer method for insoluble silicates, Five
Grabimetric and Flame Photometric Determination of Alkalies in Cement Raw hlixtures and Kiln Dusts (A11 values corrected for blanks)
____ Sample 3-0. 12724 12725 12726 12727 12728 12733 12734 12735 12736 12737 'I
Material Raw mix R a w mix R a w mix Raw mix R a w mix Kiln dust Kiln dust Kiln dust Kiln duet Kiln dust
Grav. Pt cruc. 0 30 0 22 0 03 0 22 0 21 0 6G 0 43 0 11 0 27 0 35
F.P. Pt cruc. 0.29 0.23 0.03 0.22 0.22 0 62 0.41 0.08 0.24 0.33
NaeO, 70 Diff. from F.P. J.L.S. Diff. from gray. criic.Y grav. - 0 01 0 30 0 + 0 01 0 24 +KO2 0 0 02 -0.01 0 0 24 f0.02 0 23 +0.02 +o 01 - 0 04 0 61 -0.05 - 0 02 0 41 -0.02 -0 03 0 09 -0.02 - 0 03 0 24 -0.03 - 0 02 0 37 +0.02
Single determinatlons. All other values are average of 2 or more determinations.
Grav. Pt cruc. 0.93 0 82 0 22 0 73 0 73 3 82 3 15 1 02 3 36 2 06
F.P. Pt cruc. 0.93 0.81 0.23 0.71 0.71 3 83 3.15 0.99 3 35 2.01
Kz0,
470
Diff, from grar. 0 - 0 01 +o 01 -0 02 - 0 02 +o 01 0 - 0 03 - 0 01 - 0 0.5
F.P. J.:.S. cruc. 0 0 0 0 0 3 3 1 3 2
95 83 24 76 73 75 10 00 32 08
Diff. from urav. f O 02 fO
01 +0.02 4-0 03 0 - 0 07 - 0 05 -0.02 -0.04 f O 02
ANALYTICAL CHEMISTRY
1580 Table 11.
Gravinietric and Flame Photometric Determination of -4IkaIies in Cement (.ill values corrected for blanks) SaaO. % . .______ Diff. froiri I'.P. .kcids Diff. froin Grav. Pt grav. soh. grar. cruc. 0 0.24c 0 57" t o 01 -0 02 0 19 0.02 0.03 -0 03 1 30 0.15 0 13 +o 03 0.61 +o 02 0 14 1.06 0 08 T O 06
-~
~~~
(:ray. Pt
S-unirpleS o . 1 I668
C'enient CI'IIC. Check 1 0 23" Long-'rim= Struiy 0 04 14 0 06 Long-Time Stud>18 Long-Time Studv n 12 23 Long-Time Stud;. 0.59 43 1 00 Long-Tiiiie Stud;. .iverage of 28 deterininations. Iverage of 11 deterininations. Average of 22 determinations. .ill others average of 2 t o 4 deterininationi. 13
'I '2
F.P. Pt cruc. 0 236
Sational Bureau of Standards silicates were analyzed; four by both the gravimetric and flame photometer methods and oiic by the flame photometer method only. The calcium carbonate used in the tests was J. T. Baker (:hemical Co. or Mallinckrodt Chemical Works special low in alkalies. Blank determinations of potassium oyide by the procedure described above showed practically none in either reagent,. Biniil:tr determinations of sodium oxide showed an apparent content of as much as 0.016% sodium oxide in the calcium carbonatc. This is equivalent to about' 0.025% sulfate (SO,), which is only slightly higher than the manufacturer's claims of a maximum alkali content of 0.01 to 0.02% as sulfate. Diamond and Bean (6) have reported t'hat at, the m v e length used for the photometric determinations of sodium, some iiiterference is caused by calcium prcserit in the solutions. However, all test solutions-standards (including zero), samples. and blanks-contained equal amoun t,s of calcium carbonate (containing a slight amount of sodium oxide), thus compensating for the effect of calciuni in the solutions. The addit,ional amount of carbonate used in sintering t,he s:miples and blanks was reniovecl before test, hut the sodium ositlc which it contained as :in impurity remained and caused the o1)~ervctlblank values. As about j . 0 grams of calcium carbonat,e were used in the sintering proccdure, the blanks shoiveti an appareut sodium oxide content equivalent to as much as 0.08% of the silicatc on thr basis of a 1-gram sample in a volume of 100 nil. of solution. Hence all results reported later were corrected by blank tests. For larger silicate samples in the snmc volume t,he blank \voultl have a proportionately loner value. The platinum crucibles used for Piiitering in most of t,he tests were Baker & Co. Figure Xo. 4 crucibles. 40-ml. cspacit!., with well-fitting covers. These crucibles are referred to hereafter as platinum crucibles. Some comparative tests were made with J. Lawrence Smith type platinum crucibles. referred to as
F.P.Pt criic. 0 i8h
K?O. ' C _~-____ DiR. f r o i n F.P, Arid Diff. from era\-. *oh. yrav. - 0 01 0 6OC A 0 03 0 20 +o 01 1 40 +o 10 0 14 +o 01 0 14 0 10
0
T O 02
J. L. 8. crucibles, and some with nickel crucibles similar to those used by the Bureau of Mines (11). )fan!. tests were made n i t h varying amounts of calcium c,nrbonate and ammonium chloride. vari:Ltioiis in the manner and temperature of sintering, and the use of the different types of crucibles referred to above. The best results were obtained b>the procedure given above. Samples were prepared for sintering by grinding the mixt,ures of ingredients with a power driven mort;tr arid pestle. DISCUSSION OF RESULTS
The rarv mixtures wcre analyzctl iii duplicate or triplicate u ~ i n g plat,inum crucibles for t,he sintering process. The alkali values olitained by bot,h the A S T l I gravimcti,ic :ind the proposed flame photometer method of analysis of the sintered material are shown in Table I, columns 3, 4, 5, ant1 9. The close agreement of t,he flame photoniet,er with the gravimetric method values is apparent from the small differences sho\\-n in columns 5 and 10. The alkali values for cement 11 1568. anulj.zeti in various ways arc .;ho\vn in Table 11. .is this cenieiit is the laboratory check sample. it relatively large number of tcsta have been made by nll three method?. The values sho\v t h a t the proposed method gave itlrntical or nearly identical results to those obtained bjeither of the A%STJI procrdures. The a1k:iIi v:ilues for five ot,her cements used in the 1,ong-Time Study ot' Ccmrrit Performance in Concrete. previously anal!.zed I)>. hot,li IYTYI methods, are also showi in Table 11. These values are int.roduced to show the closc agreement, between alkali values by the two .\STM nicthotls. one of which specifies the pliotonietric analysis of an acid solution of the sample. and thr. other specifies a gravimetric :i~ialysisincluding a sintering proc.css siniikir t o that used in the Iiroposed procrdure. The results of tests of five Sational I3urcSau of Standards samples of silicates are shown in Table TII. The results for potas-
Table 111. Gravimetric and Flame Photometer Determination of 2ilkalies in National Bureau of Standards Silicate Samples and a Cement (Different t y p m of crucibles for sintering.
Sniiiple so.
la 70 91 93 98 11668
Material B r g . lime Feldspar Opal lass High-%oron glass Plastic clay Cement
Grarimetrir. Pt Crucihler XBS S o . of .41k., % tests Alk., % Diff.. ''; 0.39 2.38 8.48 4.16 0.28
...
7 2 3
0.30 2.22 7.71
2 28
0.27 0.23
..
...
-n.og -0.16
-0.77
...
-0.01
All values corrected for blanks) Flame Photometer Si Criir,iblee __ Pt Crucibles
so. of trsts
.ilk., cO
Roiliiini Oxide 1 .i 0.28 4 2.26 10 7.67 8 3.48 G 0.21 11 0.23
\---
1)iB..
(;'
tests
~-
. i l k , , b' 0.31 2.08 7.83 3.60 0.24 0.26
-0.11
-0.12 -0.81 -0.68 -0.07
Diff., '1
_ _J . L .
so.of tests
-0.08 -0.30 -0.65 -0.56 -0.08
...
-
-
3 . Pt Crucibles - ~-
Alk.. %
Diff.. 'C
0.29 1.16 8.01 3.91 0.23 0.22
-0.10 -0.22 -0.47 -0.25 -0.05
11.74 12.55 3.26 0.16 0 3.6 00 7
+0.03 -0.03
...
Potassiiiiii Oxide
la 70 91 93 98 11668
.4rg. lime
Feldspar Opal glass High-boron glass Plastic clay Cement
0.71 12.58 3.25 0.16 3.17
...
7 2 3
0.71 12.24 3.27
0
15
0.il
-0.34 +0.02
io
4
12.28 3.18
2 28
3.04 0.57
-0.13
fi
2.92 0.58
..
...
...
...
8
11
0.14
0 -0.30 -0.07 -0.02
-n.2:
...
3 2 3 3
z
2
0.72 14.16 3.23
n
13 3.04
0.56
+0.01
-0.42 -0.02 -0.03
-0.13
...
3 2 4 3 2R
+on1 n
- 0. ... 1 ~
V O L U M E 26, NO. 10, O C T O B E R 1 9 5 4 sium oxide using platinum crucibles followed by either gravimetric or flame photometric measurements are in good agreement with the bureau’s values. However, the sodium oxide values, with the exception of those for No. 70, feldspar, are consistently low. Many te& i%-eremade, as shown in Table 111, using various modifications of the methods in an effort t o duplicate the bureau’s values more closely, but without success. Believing that the use of the regular platinum crucibles may have been a source of error, the tests were repeated with nickel crucibles similar to those used by the Bureau of Mines ( I I ) , or standard J. Lawrence Smith platinum crucibles. The sodium oside results, as shown in Table 111, were still low for four out of five of the samples, although the J. L. S. crucibles gave soniewhat closer values than either of the other kinds of crucibles. The reasons for the discrepancies are not entirely clear. The errors due to the presence of sodium oxide in the calcium carbonate used in sintering the samples were over and above the compensating errors caused by the calcium (and traces of sodium) introduced in equal amounts into both the standard (calibration) solutions and solutions under t,est. I n the case of gravimetric analyses, however, there is no compensation for impurities in the calcium carbonate. Hence, all the author’s reported values were corrected for blank tests. However, if these corrections were not applied, the values would be much closer to the values reported by the Xational Bureau of Standards and by other investigators who have analyzed some of the same standard samples by flame photometric methods. For example, the values shown in Table I11 for samplc l a , argillaceous limestone, uncorrected for blanks, would range from 0.36 to 0.39% sodium oxide, in close agreement. with the certified value of 0.39%. With sample la, an examination of the Bureau of Standards’ certificate of analysis showed that t,he material was analyzed over 20 years ago and that the values used in computing the certified average range from 0.31 to 0.45% sodium oxide. The J. L. S. crucibles, having shown somewhat higher sodium oxide values than other types of crucibles on the t,ests of the standard samples, were used for repeat tests of the ratv mix and kiln dust samples (columns 6 and 11 of Table I). The differences,
1581 hoivever, as shown in coluninh T xritl 12, were neither sigiiificant nor consistently in the same direction. Therefore, considering all the data in Table I it appears that the results &tainecl by the proposed flame photometric method, wing either regular platinum or J. L. S. crucibles, are in good agreement with the longer gravimetric procedure. TIME REQUIREMENTS
The time required for allCali analysis hy t,hc proposed met,hotl was less than half of t,hat required for gravimetric analysis of acid-insoluble silicates. One uf t,hr primary objects of this study has been to get an accur:tt8tna? \ v ~ 1 1as a rapid method. It ie possible that for less accurate, rout,ine work the method would rrquire one third or 1w1; of thc. time of t h e gravimetric nictliod. \\CKNOWLED(:MENT
The author tvishes t,o wcliiion.lc:tIge the contributions of Helen
K. McJTillen and William G. IIime in connection with certain phases of the analytical study.
niitl
literature rrsearoh portions of this
LITERATURE CITED
Testing Materials, I’hiladelphia. “.iBTJI Standards Cement (With Ltelatcd Information) ,” Designation C 114-47, Sec. 2 1 . 2 2 , up. 70 2 , . t p d 1952. ( 2 ) Ihid., Designation C228-4!)T, ~ I J . 157 ~ 6 1 . (3) Becker, F., Zement-Kulk-Gips. 4 , 93-7 (1951). (4)Bi,ffen, F. MI., ~ A L CHEY.. . 22, 1014- 17 (1950). (5) Dialnolid, J. J., and Ikan. 1,conard. I h i d . , 25, 1825-30 (195.7). (0) Euhaiik, W. It., and Boguc. I t . I f . , .J. Research A‘atl. B U T . Sta?idurds, 43, 173--81 (1941)).
(1 ) .Ini. Soc.
on
iV.,Jr.. Soil S c i . , 71, 211 14 (1951). W.J., arid Chaikcn, I
